How to implement AI in your business Me pēhea te whakatinana i te AI ki tō pakihi

The first step to a successful AI trial is choosing when and where to start. Create a list of your pain points – these will become your focus areas. 

Think about the day-to-day running of your business and ask yourself: 

• What takes longer than it should? 
• What feels repetitive? 
• What could be automated?

Write down 5–10 answers to these questions. Then pick one answer that:

• has low costs associated with it
• has low risks associated with it
• if improved, could have a positive impact on your business.

This is your focus area for your pilot. 

Do your research and choose a general AI tool for your test — for example, one that can help with writing, summarising, or generating ideas. Tools such as ChatGPT, Claude, Gemini or Copilot can be good starting points.

Check whether the applications you use every day to do your work already include AI features – for example, Microsoft 365 or Google Workspace. Starting with AI that’s built into your existing systems can make your test faster and easier to set up.

If you need to use tools that aren’t built into your existing apps, choose AI tools with business‑grade paid plans, as these usually offer stronger administrative controls and clearer information protection. Review each product’s data and privacy settings. If you can, switch off any options that allow your prompts or results to be used by the AI company’s developers for tool improvement or training. 

Select a simple, low-risk task within your chosen focus area to complete using AI. A good task to start with is generally one that's repetitive or time consuming, has an task result that’s easy to review, and where the impact of a mistake is low.  

Examples include:

• rewriting an email to be clearer, more professional, or friendlier
• drafting a quote or simple proposal
• drafting a social media post or job advert
• summarising meeting notes and action items. 

Once you’ve completed the task, check the results are accurate and align with what you asked for. You are still responsible for the final result, so checking it holds up to scrutiny is important.

Take time to compare how the task was delivered using AI compared to your normal business processes. 

When doing the comparison, record any changes you noticed. For example: 

• Time – did it take more or less time to complete the task using AI? 
• Quality – was the quality of the result lower, similar or better? 
• Risk – were there any obvious risks with privacy, tone or accuracy that need to be mitigated?

Think about how this could affect different tasks in your business that differ in type, size or complexity. This will help if you choose to progress beyond the pilot stage. 

For sole traders, this is easy – you’re responsible. For businesses that have more than a handful of employees, the first step to implementing AI on a more permanent basis is nominating a contact person in your business.  
 
Have someone act as an owner or manager for implementing AI in your business. This helps keep decision-making clear, and reduces the chances of different staff using any kind of AI tools for their work (also called shadow AI) that haven’t been approved or with no oversight. 
 
Choose one person to manage access to AI tools – including approving tools, assigning or removing licences, and turning connectors on or off. 

To successfully and safely embed AI in your business, you must have agreed rules and processes in place for using it.  
 
Make sure you have policies and processes that cover:

• what approved AI tools staff can use. 
• what information must not be entered into AI tools – for example, personally identifiable information from customers or staff.  
• when and how a human must review AI outputs – for example, to check specific financial, legal, and safety tasks, or that customer communications are on brand.
• how you ensure the quality and accessibility of information for AI tools – for example, by keeping sales and inventory data clean and organised.  
• how and where to keep a record of AI outputs, so these don’t get lost or deleted.
• how to use agentic AI safely – for example, keep permissions tight and have a person confirm important steps. Watch for prompt‑injection risks – where malicious content in a webpage or file tricks the AI tool into taking unintended actions – when tools browse the web or interact with other software.

You may need to create new policies and processes to cover AI or update existing ones, including IT, social media and privacy policies.  

It’s important to have a process for if something goes wrong. You’ll need to stop the activity, contain the issue, assess what was shared, fix the process or settings, and communicate with anyone who needs to know. Recording the incident and any follow‑ups helps prevent the same issue happening again. 

Keep a simple “AI use log” in a shared place for any higher‑risk tasks. Record the date, who used the tool, what they did, what inputs were used, who reviewed it, where the final version is saved, and any issues. 

Support your team to learn more about AI and build the foundational skills they need to use it. The more your staff experiment with AI tools, the easier it gets. Focus on these key skills first, before encouraging experienced and interested staff to learn more.  
 
Encourage your team to embed their learning by providing safe, approved AI tools, and clear guidance and processes. Explain to staff why using personal or free AI accounts at work has security and privacy risks. When staff use AI tools at work that aren’t approved, this is often referred to as shadow AI. 

Choose one area of your business to implement AI first. For example, human resources or marketing. Improve the end-to-end processes and workflows in this area.  
 
When you see a real improvement, take your learnings and apply these to other areas of your business. Expand step-by-step and based on evidence. 
 
Once you’re comfortable with your operational implementation, think bigger. Look for opportunities in your business to implement AI on a systemic or strategic level.